The plastic optical fibers have excelent physical properties

When transmitting light from a spatially and temporarily incoherent source (i.e. not a laser), then you simply need the largest core area and the largest numerical aperture (collecting angle). Plastic fibres are good because they are still flexible for large diameters.

I assume your intention is to focus sunlight onto the fibres, and arrange for the parabolic mirror to actively track the sun.

Which wavelengths of light are interest? Presumably visible wavelengths suitable for photosynthesis - predominantly 400 nm to 700 nm, with the emphasis on blue and red wavelengths.  http://hyperphysics.phy-astr.gsu.edu/hbase/Biology/ligabs.html

How far do you need to transmit the light?

Polymer core optical fibres are attractive, as they are available with large core sizes, high numerical aperture, and are quite flexible.  Attenuation is typically less than 0.4 dB/m for some red light, and can be lower.  Transmission might be 90% over 1 m, but only 60% over 5 m. http://www.ieee802.org/3/GEPOFSG/public/Sep_2014/Tsukamoto_GEPOF_01b_0914.pdf

A rather older review of POF by Chris Emslie from 1987 has some useful background: http://www.orc.org.uk/publications/03xx/302.pdf

Silica core fibres have low absorption losses throughout the visible spectrum, and can tolerate higher operating temperatures and sustained exposure to concentrated ultra-violet light.  A disadvantage is that the larger core diameters are less flexible than polymer fibres. If you don't need NIR (near infra-red) wavelengths, choose a high-OH silica.  Absorption can be less than 0.07 dB/m from 400 nm to 870 nm, so transmission better than 90% over 5 m fibre.  https://www.thorlabs.com/navigation.cfm?guide_id=2284

https://www.thorlabs.com/newgrouppage9.cfm?objectgroup_id=362

What concentration factor do you expect from your parabolic mirror?  An upper limit is set by the focal ratio of the mirror, and by the numerical aperture of the light accepted by the fibres. 

The sun subtends an angle of 0.53 degrees.  A numerical aperture of 0.47 (f/1.08) corresponds to a theoretical concentration factor of 10000.  The available solar radiation (solar constant) is around 1 kW m-2, so irradiance at the focus of the mirror could be as high as 1 kW cm-2.  This is not an intrinsic problem for an all-glass silica fibre, but absorption of infra-red radiation by polymer cladding, protective coating or polymer jacket could result in a significant temperature rise, with potential for damage or longer term degradation..

Thermal loading will be lower at larger focal ratio (f-number), and the manufacture of the mirror will be simpler.  At f/2.5 (NA=0.2), all-silica fibres are available with a polyimide coating, suitable for operation up to 300 C, and with improved resistance to solarisation - though UV damage should not be a major problem with good quality high-OH synthetic silica.

If your concentrator is relatively small, then Thorlabs offer ready-made fibre bundles https://www.thorlabs.com/newgrouppage9.cfm?objectgroup_id=5609

For higher throughput, a hollow waveguide with a highly reflective inner coating could be more cost effective - possible at the expense of more complex optics to couple light from a tracking parabolic concentrator.

http://www.renewableenergyfocus.com/view/13225/spi2010-3m-launches-new-solar-mirror-films/

http://renewable-energy.3mil.co.il/?mod=download&table=files&field=fi_pdf&me_id=90

Thanks a lot Alan Robinson!

I want to use it for plant lighting, so a spectra of 400-700 is what i want to achieve.

I will be transfering the light around 5meters so i don't think i will have much problems with losses. Of course an optical fiber which can transfer light further is better but also more expensive so if there no significant difference at the five meters distance then i'll go with the less expensive one.

I need to do some research on the concentration factor because at the moment i use an old satelite dish so i have yet to figure out its characteristics.

A cooling system is essential and will be used in order to protect the fibers.

Thanks for the extra information on the waveguide.

If you focus on the power density and damage threshold, the hollow core fibers can be competent.

http://www.nktphotonics.com/product/hollow-core-photonic-crystal-fibers/

@Chenge Wang

As Richard Epworth pointed out, tor efficient coupling from a broad bandwidth extended incoherent source, coupled power depends on the number of modes supported by the fibre - proportional to the square of the product of core diameter and numerical aperture. 

https://en.wikipedia.org/wiki/Etendue#Conservation_of_etendue

The NKT HC-580-02 photonic crystal fibre has a core diameter of 6.6 μm, attenuation exceeding 1 dB/m outside the range 560 to 610 nm, and appears to be single mode, presumably supporting only two orthogonally polarised modes.  A reasonable choice if the source is a high power single transverse mode laser, and propagation losses of 1 dB/m are acceptable.

A 1 mm diameter polymer or silica core fibre has lower attenuation for visible wavelengths.  With a modest 0.25 NA launch, it will support in excess of 2000000 modes, and will couple of order 106 times more power than a single mode waveguide.  This seems a better fit to the present application.

Thorlabs quote 1.8 kW CW power capability for their 0.6 mm core TECS-clad fibre, and 5 kW for 1 mm core.  This should be more than adequate if thermal loading becomes a problem with polymer fibre.  https://www.thorlabs.com/newgrouppage9.cfm?objectgroup_id=6845